Add coverage summary stats and --minimum-coverage-pct filter to validate

[mmacferrin]

Summary

Adds coverage-aware reporting and filtering to ivert validate, plus a couple of small _summary_stats.txt improvements.

Coverage summary statistics

When -mc/--measure-coverage is used, _summary_stats.txt now includes a decile section reporting the RMSE of grid cells above each coverage threshold (from 100% of cells down to the best-covered 10%). This shows how ICESat-2 coverage — and thus sampling bias — affects reported DEM accuracy.

--minimum-coverage-pct / -mcp filter

New option on ivert validate that drops grid cells whose measured coverage is below a given percentage (0-100) from all outputs: the _results.h5 dataframe, _summary_stats.txt, error exports, and plots.

• Requires -mc/--measure-coverage; using it without that flag raises a clear usage error.
• Value is constrained to the 0-100 range.
• Threaded through the single-DEM, multi-DEM (collection), and DEM-subdivision code paths. Because coverage is a per-cell property, per-subset filtering yields identical merged results.

_summary_stats.txt polish

• Reduced excessive float precision to 2 decimal places (at least 2 significant digits for magnitudes below 0.10).
• Reordered so the "Number of cells with bathymetry photons" and "Mean roughness" lines appear above the "== Decile ranges of errors" section.

Testing

• All three edited modules compile; ruff check/format pass.
• CLI verified: option appears in --help, the -mc-required guard fires, and out-of-range values are rejected.
• Filtering verified end-to-end against a synthetic results dataframe (a ≥50% threshold retained exactly the expected cells).
• No existing unit tests cover this path (the src/tests/ files are integration download scripts).